AbstractThe detection of targets with a radar system is inherently a nonlinear stochastic process. Thus the application of nonlinear estimation becomes inevitable. This paper is concerned with two approaches to nonlinear estimation:an approach employing a Gaussian second-order filter (GSF) and another employing an iterative filter (IF). The accuracy of estimates obtained via such filters are compared and shown to be superior to that obtained via the extended kalman filter (EKF). Moreover, the idea of switching models has been extended to the former and results in a new scheme called variable dimension GSF (VDGSF). This latter technique is shown to yield substantially better performance. The simulation results presented validate the performance predictions of the proposed algorithm.